DNA mismatch repair is a key mutation avoidance pathway that is of clinical interest for several reasons. Inactivation of mismatch repair is the cause of a common form of hereditary colon cancer and has been implicated in the development of a subset of sporadic tumors. Mismatch repair defects also have implications for cancer therapy because inactivation of the pathway renders cells resistant to the cytotoxic effects of certain anti-tumor drugs, a consequence of participation of the system in the DNA damage response. Perhaps surprisingly, mismatch repair function is also required for the production of certain mutations, such as the expansion of (CAG)n repeat sequences, the primary cause of a number of neurodegenerative diseases. By elucidating the molecular nature of human mismatch repair, we hope to understand its roles in controlling the occurrence of mutation. To this end we propose four lines of work: (1) Available information on the nature of strand-directed human mismatch repair indicates that the course of the reaction is dictated by an evolving set of protein-protein and protein-DNA interactions, and that repair events initiated by the mismatch recognition activities MutS1 and MutS2 differ in significant ways. By analyzing the nature of selected multi-protein and multi-protein-DNA complexes, we hope to further clarify the mechanisms of MutS1- and MutS2-initiated repair events. (2) The somatic expansion stage of (CAG)n neurodegenerative diseases, which depends on the mismatch repair activities MutS2 and MutL1, can occur in postmitotic cells, suggesting involvement of repair DNA synthesis in this process. The nature of processing of (CAG)n repeat elements by the human mismatch repair system will be addressed in both extract and purified systems. (3) Mismatch repair function is required for checkpoint and apoptotic responses to SN1 DNA methylators. Using a biochemical approach, we will pursue the mechanisms of MutS1- and MutL1-dependent activation of the ATR damage-signaling kinase in response to O6-methylguanine, the primary cytotoxic lesion produced by this class of drug. (4) Collaborative studies with the laboratory of Lorena Beese will address the structural basis of lesion recognition and processing by the human mismatch recognition system.